• 

• Selection on IV
• Selection on DV

Selection on Both

• When your case study suffers from selection bias,
  
  

  

• Any characteristics that the selected cases share is a cause.
• The relations between variables within the selected set reflect a general pattern.

• Question: Why new industrializing countries (NICs) grows more rapidly?
• Theory: The repression of labor

• Case:

  

• What type of design this is?

• Theory-oriented
  • Labor repression contributes to growth in a at least mixed economy
  • Case selection: Communist countries are excluded.
• Hypothesis-oriented
  • Theory: Military governments are more likely to negotiate their extrication from power than are personalist regimes.
  • Hypothesis: Incidence of negotiation in military governments during the years in which breakdown occurs is higher than that in personalist regimes.
  • Case selection: Only breakdown regimes are select.

• Being clear about the domain of the variables
• Coding sheet

• "What was the effect of cause X?" \(\checkmark\)
• "What cause Y?" or "Variance of the effect of X?" \(\times\)

• If the theory can make it here, it can make it anywhere.

• E.g.: Evangelista(1999)
  • Target: Influence of transnational actor
  • Theory: Transnational actor can affect international relations
  • Case: U.S. vs. USSR
    • Scientists' contact affect the course of U.S. and Soviet defense and arm control policies.

• Control similarity, focus on differences.

• E.g.: Ray(1995)
  • Target: Interstate conflicts
  • Theory: Democratic peace
  • Case: UK-France Fashoda Crisis vs. Spanish-American War
    • Control for confounders (regression effects, mortality, selection bias) and same year
    • Explanatory: regime

• Cases that do not conform to the predictions made by the theory or theories under investigation.

• E.g.: Elman(1997)
  • Theory: Democratic peace
  • Case: Finland vs. UK in the Continuation War
    • Finland: Legislature does not balance president
  • Conclusion: decentralized vs. centralized democracies

• Explicit attention to and process tracing on alternative explanations
• Sustained focus on the question of "what else must be true" of the process
• A wide variety of sources